The AMI Meeting Corpus consists of 100 hours of face-to-face meeting recordings; the associated AMIDA Meeting Corpus consists of 10 hours of meeting recordings where one participant is connecting to a meeting room using desktop conferencing. The meetings were recorded in English using three different rooms with different acoustic properties, and include mostly non-native speakers. Some meetings make use of an elicitation technique in which the participants play roles in a design team; others use other elicitations or are completely naturally occurring. Large portions of the corpus have been annotated for a wide range of behavioural phenomena.
The corpus has been extended by SSPnet to include manual annotations for social role and subjectivity, and these are included in the latest public release of the AMI corpus, along with a number of automatic annotations computed over both manual and ASR-generated transcripts.
- edition: 1.4, 1.0
- url: http://corpus.amiproject.org/
- main_author: The AMI Consortium
- license: Creative Commons Non Commercial Attribution Share-Alike
- subjects: 189 in completely face-to-face trials; 24 in remote trials
- recordings: 175 in completely face-to-face trials; 22 in remote trials
- duration: 30 minutes
- naturality: roleplay
- media: synchronized close-talking and far-field microphones, individual and room-view video cameras, and output from a slide projector and an electronic whiteboard; unsynchronized pen outputs
- language: English
- interaction: group
- annotation: many different dialogue and group interaction aspects
Categories: multimodal-analysis
Copyright © 2012